English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Training accurate deep neural networks (DNNs) in the presence of noisy labelsis an important and challenging task. Though a number of approaches have beenproposed for learning with noisy labels, many open issues remain. In thispaper, we show that DNN learning with Cross Entropy (CE) exhibits overfittingto noisy labels on some classes ("easy" classes), but more surprisingly, italso suffers from significant under learning on some other classes ("hard"classes). Intuitively, CE requires an extra term to facilitate learning of hardclasses, and more importantly, this term should be noise tolerant, so as toavoid overfitting to noisy labels. Inspired by the symmetric KL-divergence, wepropose the approach of \textbf{Symmetric cross entropy Learning} (SL),boosting CE symmetrically with a noise robust counterpart Reverse Cross Entropy(RCE). Our proposed SL approach simultaneously addresses both the underlearning and overfitting problem of CE in the presence of noisy labels. Weprovide a theoretical analysis of SL and also empirically show, on a range ofbenchmark and real-world datasets, that SL outperforms state-of-the-artmethods. We also show that SL can be easily incorporated into existing methodsin order to further enhance their performance.

Symmetric Cross Entropy for Robust Learning with Noisy Labels